Hyper-selectivity, Racial Mobility, and the Remaking of Race

Transcription

1 Hyper-selectivity, Racial Mobility, and the Remaking of Race Van C. Tran, Jennifer Lee, Oshin Khachikian, Jess Lee RSF: The Russell Sage Foundation Journal of the Social Sciences, Volume 4, Number 5, August 2018, pp (Article) Published by Russell Sage Foundation For additional information about this article Access provided at 11 Apr :41 GMT with no institutional affiliation

2 Hyper-selectivity, Racial Mobility, and the Remaking of Race Van C. Tran, Jennifer Lee, Oshin KhacHikian, and Jess Lee Recent immigrants to the United States are diverse with regard to selectivity. Hyper- selectivity refers to a dual positive selectivity in which immigrants are more likely to have graduated from college than nonmigrants in sending countries and the host population in the United States. This article addresses two questions. First, how does hyper- selectivity affect second- generation educational outcomes? Second, how does second- generation mobility change the cognitive construction of racial categories? It shows how hyperselectivity among Chinese immigrants results in positive second- generation educational outcomes and racial mobility for Asian Americans. It also raises the question of whether hyper- selectivity operates similarly for non- Asian groups. While there is a second- generation advantage among hyper- selected groups, hyper- selectivity has not changed the cognitive construction of race for blacks and Latinos as it has for Asians. Keywords: hyper- selectivity, racial mobility, assimilation, second generation, racial categories, identity Today s immigrants have more diverse national origins than ever before in U.S. history. As a result, race and immigration have become inextricably linked in the United States; one can no longer understand the complexities of race without considering immigration; correlatively, one cannot fully grasp the debates in immigration without considering the role of race in U.S. society. Immigrants are diverse with respect not only to national origin, but also to selectivity. At one end of the extreme are Asian Indians, Chinese, Nigerians, Cubans, and Armenians who are, on average, hyper- selected; not only are they more likely to have graduated from college than their nonmigrant counterparts, but also more likely to have a college degree relative to the U.S. mean. At the other end of the extreme are groups such as Mexicans who are hypo- selected, that is, less likely to have graduated from college than their nonmigrant counterparts and the U.S. mean. At 28 percent of the foreign- born population, Van C. Tran is assistant professor of sociology in the Department of Sociology at Columbia University. Jennifer Lee is professor of sociology in the Department of Sociology at Columbia University. Oshin Khachikian is a PhD candidate in sociology at the University of California, Irvine. Jess Lee is a PhD candidate in sociology at the University of California, Irvine Russell Sage Foundation. Tran, Van C., Jennifer Lee, Oshin Khachikian, and Jess Lee Hyper- selectivity, Racial Mobility, and the Remaking of Race. RSF: The Russell Sage Foundation Journal of the Social Sciences 4(5): DOI: /RSF Direct correspondence to: Van C. Tran at Department of Sociology, Columbia University, 607 Knox Hall, 606 W. 122nd St., New York, NY Open Access Policy: RSF: The Russell Sage Foundation Journal of the Social Sciences is an open access journal. This article is published under a Creative Commons Attribution- NonCommercial- NoDerivs 3.0 Unported License.

3 the remaking of race 189 Mexicans are by far the largest immigrant group in the country and one of the most socioeconomically disadvantaged. Their sheer size, coupled with their hypo- selectivity and disadvantaged socioeconomic and political status, have placed them at the center of research, debates, and policy prescriptions about immigrant assimilation and comprehensive immigration reform. By comparison, relatively little attention has focused on the assimilation patterns of hyper- selected immigrant groups such as the Chinese and Asian Indians, even though China and India have passed Mexico as the top sending countries for immigrants to the United States since In this article, we shift the focus to hyperselected immigrant groups, and ask how they may be changing our cognitive construction of U.S. racial categories in the twenty- first century. First, how does hyper- selectivity affect the educational outcomes of the second generation? Second, how have the achievements of hyperselected immigrant groups and their secondgeneration children changed the cognitive construction of race? We tackle these questions by focusing on patterns of educational attainment among four hyper- selected groups Chinese, Cubans, Nigerians, and Armenians who are racialized as Asian, Hispanic, black, and white, respectively, in the U.S. context. We adopt a cognitive approach and propose that a change in the selectivity of an immigrant group can change the host society s perceptions of the immigrant group and may also affect the perceptions of the racial group to which they are assigned (Brubaker, Loveman, and Stamatov 2004; Wimmer 2008). Immigration, Diversity, and Hyper- selectivity The influx of new immigrants to the United States became possible with the passage of the Hart- Celler Act in 1965, which eliminated quotas based on national origin and opened the door to newcomers from non- European countries. This change brought such a dramatic shift in national origins of immigrants that today more than four in five hail from Latin America, Asia, Africa, or the Caribbean, and only one in seven from Europe or Canada (Lee and Bean 2010). The shift is the single most distinctive feature of the country s new immigration. The change in the national origins of today s newcomers has made an indelible imprint on the nation s ethnoracial landscape, transforming it from a largely black- white society at the end of World War II to a kaleidoscope of ethnoracial groups (Alba and Nee 2003; Alba and Foner 2015; Foner and Fredrickson 2004; Waters, Ueda, and Marrow 2007). Since 1965, Latinos and Asians have more than quadrupled in size from 4 and 1 percent of the population to 18 and 6 percent, respectively. Latinos are now the largest minority group, and Asians the fastest growing group (Lee and Zhou 2015; Wong et al. 2011). Driving the growth of the Asian population is immigration; 65 percent of U.S. Asians are foreign born, a figure that increases to 80 percent among Asian adults. Among Latinos, 35 percent are foreign born. Although the total black population increased by only 1 percent (from 11 to 12 percent) since 1965, the foreign- born proportion grew to 10 percent of the total U.S. black population, up from 1 percent. The group that has decreased in size since 1965 is non- Hispanic whites. Although they remain by far the largest group in the country, accounting for some 65 percent of the population, their proportion has steadily declined since 1970, when the figure was 80 percent. National origin and ethnoracial diversity are only two dimensions of contemporary immigrant diversity. Today s newcomers are also diverse with respect to socioeconomic status, legal status, selectivity, and phenotype all of which affect patterns of immigrant and secondgeneration integration. For example, Asian Indians, Chinese, Koreans, Cubans, Nigerians, and Armenians are hyper- selected. Their positive selectivity places them and their U.S.- born children at a more favorable starting point in their quest for socioeconomic attainment compared to other second- generation groups, and even compared to third- and higher- generation whites and blacks. At the other extreme are Mexicans, who are hypo- selected. Their negative selectivity, coupled with the lack of legal status, places Mexican immigrants and their second- generation children at a disadvantaged starting point

4 190 Immigration and Changing Identities (Bean, Brown, and Bachmeier 2015). Although their second- generation children make enormous intergenerational strides, they remain below the U.S. mean with respect to educational attainment (Lee and Zhou 2015; Telles and Ortiz 2008; Tran and Valdez 2017). Hyper- and hypo- selectivity have cultural, institutional, and social psychological consequences for the educational attainment of the second generation (Lee and Zhou 2017, 2015). The hyper- selectivity of Chinese immigrants can enhance the educational outcomes of the second generation, even among those from working- class families in ways that defy the classic status attainment model. For example, Chinese immigrants who arrive with more education and socioeconomic resources create ethnic capital in the form of supplemental education programs, SAT prep courses, and tutoring services that are accessible to working- class coethnics (see also Kasinitz et al. 2008; Tran 2016). Moreover, the high achievers become the role models and mobility prototypes to which group members aspire, and the reference group against whom they measure their success. These coethnic resources and cross- class social ties give second- generation Chinese including those from working- class backgrounds a leg up over other groups. In addition, hyper- selectivity has social psychological consequences, which affect in- group and out- group perceptions. For example, the hyper- selectivity of Chinese immigrants drives the perception that all Chinese are highly educated, smart, hardworking, and deserving (Lee and Zhou 2015). And, critically, because of the racialization process that occurs in the United States, perceptions of Chinese extend to other Asian immigrant groups such as Vietnamese, even though the latter are not hyper- selected. These are the spillover effects of hyperselectivity (Hsin 2016), which have resulted in the racial mobility of Asian Americans the change in status or position of a racial group (Lee 2015). Here, we draw from Aliya Saperstein s racial mobility perspective, which accounts for the shift in an individual s racial status based on changes to their social status (2015). We build on this perspective by noting that racial mobility can also occur at the group level as a result of changes in an ethnoracial group s immigrant selectivity or socioeconomic status. These changes can affect out- group perceptions, alter the group s position in the U.S. hierarchy, and lead to racial mobility for both the ethnic group as well as their proximal host racial group. This is precisely what happened in the case of U.S. Chinese and Asians. Less than a century ago, Chinese immigrants were described as illiterate, undesirable, and unassimilable foreigners, full of filth and disease, and unfit for U.S. citizenship. In 1882 Senator John F. Miller, Republican of California, told the Senate on February 28, It is a fact of history that wherever the Chinese have gone they have always taken their habits, methods, and civilization with them; and history fails to record a single example in which they have ever lost them. They remain Chinese always and everywhere; changeless, fixed and unalterable. Senator Miller added, If the Chinese could be lifted up to the level of the free American, to the adoption and enjoyment of American civilization, the case would be better; but this cannot be done, he concluded. Forty centuries of Chinese life has made the Chinaman what he is. An eternity of years cannot make him such a man as the Anglo- Saxon (see Dunlap 2017, A2). As marginal members of the human race, they were denied the right to naturalize, denied the right to intermarry, residentially segregated in crowded ethnic enclaves, and legally barred from entering the United States for ten years beginning in 1882 with the passage of the Chinese Exclusion Act (Okihiro 1994; Takaki 1979). Despite decades of institutional discrimination, racial prejudice, and legal exclusion, Chinese have become one of the most highly educated U.S. groups and are now hailed as a successful group to be emulated. The change in their immigrant selectivity and more specifically their hyper- selectivity has led to the racial mobility of not only Chinese but also Asian Americans. Facilitating the group mobility of Asian Americans is that the Chinese are the largest Asian ethnic group in the United States. Although Jennifer Lee and Min Zhou illustrate how hyper- selectivity affects secondgeneration Asian- origin immigrant groups (Chinese and Vietnamese), they do not con-

5 the remaking of race 191 sider how it may operate for non- Asian immigrant groups (2015). We expand the theoretical discussion of hyper- selectivity, and consider how it affects immigrant groups such as Cubans, Nigerians, and Armenians, and their U.S. proximal hosts Latinos, blacks, and whites, respectively. 1 We posit that though the hyperselectivity of Cubans, Nigerians, and Armenians positively affects the socioeconomic outcomes of immigrants and their secondgeneration children, it does not change groupbased perceptions of their proximal hosts as it does for Asians. In other words, although hyper- selectivity has changed the cognitive construction of Chinese, and has led to the racial mobility of Asian Americans, it has not done the same for other U.S. racial groups. Instead, Cubans and Nigerians are perceived as the exceptions to Latinos and blacks a perception that these ethnic groups actively strive to maintain as they distance and identify themselves in opposition to their proximal hosts. By contrast, Armenians like European immigrant groups of the past are becoming absorbed as whites. Four Hyper-Sselected Immigrant Groups at a Glance We provide brief immigration histories of four hyper- selected groups: Chinese, Cubans, Nigerians, and Armenians that are racialized as Asian, Latino, black, and white, respectively, in the U.S. context. Chinese Since 1965, Chinese immigrants have become the most populous Asian- origin group, from 235,000 in 1960 to more than four million in 2010 (U.S. Census Bureau 2017). Although they constitute only 1.2 percent of the total U.S. population, more than half have graduated from college, making them one of largest, most visible, most educated, and upwardly mobile groups in the country. Their ascendance has captured the attention of the media, pundits, and researchers who have provided a bevy of explanations for their educational attainment the most popular of which was the essentialist cultural argument in which pundits point to unique Chinese and Asian cultural traits and values to explain their high achievement (Chua and Rubenfeld 2014). Social scientists, on the other hand, relied on the status attainment model to explain variance in socioeconomic attainment, parental education being the strongest predictor of children s educational attainment. This model explained differences between and within nativeborn whites and blacks, but it failed to account for a vexing achievement paradox. Left unanswered is how the children of Chinese immigrants whose parents have less than a high school education, and work in ethnic restaurants and factories, attain the same education (if not more) as their counterparts whose parents are college- educated professionals. Immigration researchers tackle this paradox head on. Not only do they expose the fallacy of the culturally reductionist approach, they also explain how race and ethnicity serve as resources for immigrant and second- generation groups like the Chinese (Kasinitz et al. 2008; Hsin and Xie 2014; Portes and Zhou 1993; Zhou and Kim 2006). They point to both structural advantages such as contexts of exit and reception, ethnic capital, racial phenotype, favorable out- group perceptions, and cultural repertoires of achievement that affect second- generation success. Lee and Zhou extend this literature by adding that hyper- selected immigrants import class- specific cultural institutions and practices from their countries of origin, and recreate those that have the most utility in their new host country (2015). Hence, what may be perceived and defined as the transmission of cultural traits and values is in fact class- specific in origin. In addition, they show that the chil- 1. Proximal host refers to the racial category to which the immigrants would be assigned following immigration (Mittelberg and Waters 1992, 412). Specifically, it refers to the native- born racial group in the host society that is closest to a given immigrant group. Although we use third- plus- generation whites and third- plus- generation blacks as the proximal hosts for Armenians and Nigerians, respectively, we depart from Philip Kasinitz and his colleagues and use third- plus- generation Latinos, rather than Puerto Ricans, as the proximal host for Cubans (2008). Finally, we add third- plus- generation Asians as the proximal host for Chinese.

6 192 Immigration and Changing Identities dren of hyper- selected groups benefit from social psychological processes. For example, because Chinese immigrants are hyper- selected, teachers perceive all Chinese students as smart, hardworking, disciplined, and deserving. This can lead to stereotype promise being viewed through the lens of a positive stereotype that can boost performance. Because of the racialization process in the United States, the hyperselectivity of the Chinese extends to other East Asian groups, such as the Vietnamese. Hence, even mediocre second- generation Chinese and Vietnamese students gain advantages and second chances in the domain of education that are denied to other groups, including nativeborn whites. In turn, these cumulative advantages can result in a self- fulfilling prophecy of high achievement among Asian Americans. Van Tran adds to this body of research by clarifying that not only does hyper- selectivity matter, but so does the socioeconomic diversity of the coethnic community (2016). Although Chinese immigrants are hyperselected, the range of human capital attributes within the ethnic group is unusually wide. Thus, Chinese social networks serve to link poor and working- class people to uppermiddle- class professionals more often than in other ethnic groups, providing working- class and working- poor Chinese immigrant parents with access to cultural knowledge often reserved for upper- middle- class professionals. These direct and indirect connections through ethnic social networks facilitate the transfer of practical knowledge of the strategies necessary for educational mobility from magnet public high schools entrance exams to prerequisites for successful applications to the most selective universities. Furthermore, Tran finds that second- generation Chinese from working- class backgrounds strive to excel in school in order to obviate the prejudice experienced by their immigrant parents, and to repay them for the hardship that they have had to endure in their new host society. Thus, the superior academic credentials and socioeconomic characteristics of the secondgeneration Chinese result from the hyperselectivity of their immigrant parents, its spillover effects, and the socioeconomic diversity of Chinese Americans. These structural and social psychological advantages create ethnicspecific cross- class opportunities beyond the parental home for both middle- and workingclass coethnics as the second generation come of age. Cubans More than 2.1 million Americans identify as Cuban and, like the Chinese, are hyper- selected (U.S. Census Bureau 2017). Among initial waves of the post migrants from Cuba, 33 percent had earned a college degree, relative to only 1 percent of the Cuban national population (Pedraza- Bailey 1985). This early form of hyper- selectivity was driven by the Cuban revolution, which dislodged the dominant social classes from their homeland and resettled them in Miami (Pérez 1986), leading researchers to call the first mass migration of Cuban elite to United States the Golden Exile (Portes 1969). As push factors in Cuba intensified and incentivized emigration, successive waves of coethnics characterized by lower levels of education and professional qualifications arrived and populated the Cuban enclave in Miami (Portes, Clark, and Bach 1977; Portes and Böröcz 1989). Upon their arrival, the later waves were welcomed by a resource- rich ethnic enclave that facilitated their socioeconomic incorporation. The top- heavy class structure of the initial wave of Cuban migrants concentrated social and economic capital that would later cascade throughout the enclave and provide less- skilled coethnics with employment in the enclave (Portes and Bach 1985; Portes and Puhrmann 2015). Although the ethnic capital among Cubans in Miami has aided the socioeconomic incorporation among the first generation, the effects beyond educational aspirations are less clear among the second generation. Hyper- selectivity and socioeconomic diversity among first- generation Cubans has led to graduate degree aspirations among the second, even among the children of later wave Cuban migrants whose parents are far less likely to have graduated from college (Feliciano 2006; Rumbaut and Portes 2001). However, evidence of cross- class learning that would bolster second- generation educational attainment which is present among the Chinese has yet

7 the remaking of race 193 to be empirically documented (Haller, Portes, and Lynch 2011). Rather recent studies demonstrate that, unlike low- SES (socioeconomic status) secondgeneration Chinese who converge with high- SES coethnics in educational achievement, second- generation Cubans follow the pattern predicted by the status attainment model. Parental class predicts children s outcomes among second- generation Cubans, as reflected in the high college rates among the middle class at one extreme and high school dropout rates among the working class on the other (Portes and Puhrmann 2015). The favorable mode of incorporation and especially their context of reception has aided first- generation Cubans and has prevented downward assimilation among the second generation (Fernández- Kelly and Konczal 2005; Portes and Fernández- Kelly 2008; Portes and MacLeod 1996). Nigerians Numbering some 367,000, Nigerians make up less than 1 percent of the U.S. population, yet nearly two- thirds (62 percent) of Nigerian immigrants are college educated far exceeding the U.S. mean at 28 percent. Nigerian migration to the United States began en masse following the political upheaval in Nigeria in the 1960s, increasing rapidly through the 1990s (Ogbaa 2003; Imoagene 2012). In this decade, larger proportions of graduate degree holders and highly skilled professionals continued to flee the economic and political uncertainty in Nigeria by resettling in the United States. This more recent, hyper- selected migration converged in three U.S. cities New York, Houston, and Washington, D.C. and contributed to the growing black middle class (Logan and Deane 2003). Many of the most popular and active organizations among Nigerian Americans are not freestanding community associations created in the U.S. context, but rather American branches of hometown associations and community- based organizations with a long service history in Nigeria. Like the members of the Golden Exile who recreated Cuban private schools in Miami to ensure that Cuban parents would retain sustained authority over American- born children, Nigerians have founded mutual- aid associations across the United States that organize chain migration, assist with job placement, and direct remittances to the homeland (Konadu- Agyemang, Takyi, and Arthur 2006; Arthur 2000). The tight ethnic networks that emerge from mutual- aid associations have consequences for nonmigrants, as well as for both first- and second- generation Nigerians. For example, Onoso Imoagene reveals how these networks sustain cultural norms of advanced educational attainment among U.S. Nigerians such that they believe that it is un- Nigerian not to go to college (2017). In fact, the educational expectations among the second generation is a graduate degree, similar to that of second- generation Chinese (Imoagene 2017; Lee and Zhou 2017, 2016, 2015). Although Nigerian immigrants and their children may reduce achievement to their ethnicity, Imoagene shows how the hyperselectivity of the first generation affects the educational aspirations and attainment of the second generation (2017). Armenians Numbering approximately 460,000, Armenian Americans make up less than 1 percent of the total U.S. population, yet 44 percent of them have a bachelor s degree or higher (U.S. Census Bureau 2017). The earliest migrants fled in response to political violence and genocide, settled on the Eastern seaboard in the late 1800s (Bakalian 1992), and later moved to California to work in agriculture (Sabagh, Bozorgmehr, and Der- Martirosian 1990). Following the change in U.S. immigration law in 1965, Armenian immigrants were hyper- selected and racially classified as white, thanks to pre Armenian immigrants who successfully petitioned federal immigration officials in the U.S. Supreme Court to be classified as white. Their petition for racial classification earned them eligibility for U.S. citizenship in the 1920s (Craver 2009). Like American Jews, who are diverse in national origin yet converge in the collective memory of the Holocaust, Armenian Americans emigrate to the United States from diverse sending countries, such as Syria, Iran, Armenia, Lebanon, Iraq, Turkey and Russia, but organize collectively for federal recognition of the

8 194 Immigration and Changing Identities Armenian genocide (Waldinger and Bozorgmehr 1996). Iranian immigrants have the highest level of college completion among these groups, and Turkish immigrants, the highest level of self- employment (Sabagh, Bozorgmehr, and Der- Martirosian 1990). Coordinated by a vocal and organized political lobby, Armenian Americans benefit from a host of professional societies, youth enrichment organizations, and nonprofit hometown associations that provide social services for both the local community as well as humanitarian relief in the Republic of Armenia (Waldinger 2015; Khachikian 2016). To our knowledge, no research has been published on the educational attainment of second- generation Armenians, making our analysis one of the first mobility snapshots for this immigrant group. Given the hyper- selectivity of the first generation, their favorable context of reception, and their white racial status in the United States, it is likely that the second generation will reproduce their parents socioeconomic advantage. The ethnic capital that highly skilled professional immigrants create and sustain in a community with a high level of ethnic concentration like Los Angeles places second- generation Armenians at a favorable starting point in their quest for attainment (Der- Martirosian 2008; Phinney, Ong, and Madden 2000; Phinney, Baumann, and Blanton 2001). Patterns of Immigrant and Second- generation Educational Attainment We provide details of our data, methods, and analyses of patterns of immigrant and secondgeneration educational attainment. Data and Methods To examine the patterns of second- generation educational attainment among Chinese, Cubans, Nigerians, and Armenians, we used pooled data from the Annual Social and Economic Supplement of the Community Population Survey (CPS ASEC) from 2008, 2010, and 2012 (Bureau of Labor Statistics 2012). The CPS ASEC is the only data source that provides nationally representative samples of secondgeneration adults in the United States. The CPS ASEC is administered by the Census Bureau through both in- person and telephone interviews every month to monitor basic trends in the population. It uses a probability sample of about sixty thousand occupied households from all fifty states and the District of Columbia. The survey design features a sampling scheme under which households are included in the survey for the first four consecutive months and excluded for the next eight, before returning again for the last four. Given this sampling design, the pooling of data from the 2008, 2010, and 2012 samples ensures the presence of non- overlapping individuals in the pooled dataset, because each of these surveys was collected two years apart. The pooled sample also ensures an adequate sample size for smaller groups such as Nigerians and Armenians. The main outcome of interest is educational attainment by ethnoracial origin and immigrant generation. Our focus is on the second generation in each of the four ethnic groups. We compared their outcomes with those of the immigrant first generation from the same ethnic groups, with the proximal host from the same racial groups, and with their secondgeneration nonethnics from the same racial group. The four proximal host racial groups include third- plus- generation individuals from the same race (that is, native- born non- Hispanic whites, non- Hispanic blacks, non- Hispanic Asians and Hispanics). These three sets of comparisons were selected to reveal the complex linkages between hyper- selectivity and intergenerational mobility that underlie the cognitive construction of racial groups in the U.S. context. The analysis is restricted to respondents age twenty- five or older, given our main outcome of interest in educational achievement. This age range also allowed us to effectively compare the first and second generation in the United States with nonmigrants in their home countries for whom data on educational attainment are available only for those older than twentyfive. Our key independent variables are ethnoracial origin and immigrant generation, operationalized based on the birthplace of the respondent and those of their parents. Those with one foreign- born parent and one nativeborn parent we classified based on the ethnic-

9 the remaking of race 195 Table 1. Educational Attainment by Ethnoracial Origin and Immigrant Generation Ethnic Group % College Graduate % Total Sample N Sample Size First generation Chinese ,196 Cuban ,868 Armenian Nigerian Second generation Chinese Cuban Armenian Nigerian Second generation Non-Chinese Asian ,902 Non-Cuban Hispanic ,003 Non-Armenian white ,426 Non-Nigerian black Third-plus generation Non-Hispanic Asian ,338 Hispanic-Latino ,730 Non-Hispanic white ,480 Non-Hispanic black ,661 Total ,660 Source: Authors compilation based on the CPS ASEC (Bureau of Labor Statistics 2012). Note: Combined sample is limited to population age twenty-five and older. ity of the foreign- born parent to ensure the largest samples of the second generation. The analyses proceeded in two stages. First, bivariate analyses provided statistical profiles for each ethnic group by ethnoracial origin. Second, multivariate logistic regression analyses examined the socioeconomic attainment of Latino ethnic groups, relative to third- plusgeneration proximal hosts or to secondgeneration nonethnic individuals from the same racial group. Because the dependent variable is dichotomous, we used logistic regressions with robust standard errors and report the odds ratios. The control variables include age, the quadratic term of age, region of the country and survey year. Because CPS ASEC pools data across three survey years, we controlled for changes over time. Region is a variable with four census categories: Northeast, Midwest, West, and South. Our analyses adjusted for the stratified survey design using appropriate final weights provided by CPS ASEC. We also present some of our findings using predicted probabilities based on the multivariate analyses in which values for control variables are held constant at the mean level. Descriptive Analyses Table 1 provides an overview of our CPS ASEC pooled sample by ethnoracial origin and immigrant generation, along with the proportion with a bachelor s degree or more. Chinese were the largest among the four groups, having the highest number of both immigrant and secondgeneration respondents, followed by Cubans. In contrast, the samples of Armenians and Ni-

10 196 Immigration and Changing Identities Figure 1. Educational Attainment by Ethnoracial Origin and Immigrant Generation Percent with Bachelor's Degree or More Chinese Cuban Armenian Nigerian First generation Second generation Nonmigrant Source: Authors compilation based on the CPS ASEC (Bureau of Labor Statistics 2012), UNESCO Institute for Statistics 2012, and Education Policy and Data Center Notes: Combined sample is limited to population age twenty-five and older. Nonmigrant data for Chinese, Cubans, and Armenians are extracted from United Nations Educational, Scientific and Cultural Organization s (UNESCO) Institute for Statistics. Nonmigrant data for Nigerians are extracted from Education Policy and Data Center. gerians were rather small, reflecting both their relative group size and recency of immigration. We also specified the second- generation nonethnics from the same racial group and their third- and higher- generation proximal host groups. These two sets of comparisons provide benchmarks for second- generation progress against the U.S. mainstream and how well the four ethnic groups performed relative to other second- generation individuals of the same racial but not the same ethnic background. These two benchmarks capture the increasing diversity of the U.S. mainstream into which the second generation assimilate because recent research has shown how the choice of reference groups to compare second- generation attainment affects the conclusion of secondgeneration progress, mobility, stagnation, or decline (Jiménez and Horowitz 2013; Kasinitz et al. 2008; Portes, Aparicio Gomez, and Haller 2016; Tran and Valdez 2017). Examining educational attainment among the four ethnic groups revealed two distinctive characteristics hyper- selectivity and intergenerational mobility. Figure 1 presents descriptive results on the proportion with a bachelor s degree or higher within each ethnic group in the United States, contrasting these proportions with the educational attainment among nonmigrants in the sending countries. Among the population age twenty- five and older, firstgeneration immigrants reported significantly higher percentages of having a bachelor s degree or higher than their nonmigrant counterparts in respective home countries. This achievement gap is most striking between Chinese nonmigrants and Chinese immigrants in the United States, but also substantial for the other three groups. Only 3.6 percent of nonmigrant Chinese reported having a college education, but 52.7 percent of immigrant Chinese held a bachelor s degree. This hyper- selectivity ratio of 17:1 between immigrant and nonmigrant means that Chinese immigrants were disproportionately well educated relative to nonmigrants. This ratio is about 8:1 for Asian Indians. This gap is also quite stark among Nigerians. Immigrant Nigerians (63.8 percent) were six times more likely than their nonmigrant counterparts to report having a bachelor s degree or more (11.5 percent). Their hyperselectivity ratio is about 6:1. Similarly, 23.5

11 the remaking of race 197 Figure 2. Educational Attainment by Ethnoracial Origin and Proximate Host Non-Hispanic Asian, 3rd+ Non-Chinese Asian, 2nd Chinese, 2nd 61.2 Hispanic, 3rd+ Non-Cuban Hispanic, 2nd Cuban, 2nd 40.6 Non-Hispanic white, 3rd+ Non-Armenian white, 2nd Armenian, 2nd 57.6 Non-Hispanic black, 3rd Non-Nigerian black, 2nd 37.7 Nigerian, 2nd Percent with Bachelor's Degree or More Source: Authors compilation based on the CPS ASEC (Bureau of Labor Statistics 2012). Note: Combined sample is limited to population age twenty-five and older. percent of immigrant Cubans reported having a college degree relative to only 14.2 percent of nonmigrant Cubans, a gap of 9 percent. Among Armenians, the corresponding gap is about 10 percent. Between the first and second generation, intergenerational mobility is clear. A significantly higher proportion of the second generation from four ethnic groups reported having a bachelor s degree or higher than their immigrant first generation. Among Chinese, this number increased from 52.7 percent to 61.2 percent. Among Cubans, 23.5 percent among firstgeneration immigrants and 40.6 among the second generation reported having a college education. Among Armenians, the increase was from 34.5 percent to 57.6 percent. Among Nigerians, patterns of mobility were similarly robust, up approximately 10 percentage points to 73.5 percent in the second generation. The overall pattern is clear: the first generation was significantly more selective than the nonmigrants in the country of origin, and the second generation reported even higher education than the first. Figure 2 presents descriptive results on the proportion with a bachelor s degree or higher for the second generation from the same four groups, contrasting these proportions with second- generation nonethnic counterparts

12 198 Immigration and Changing Identities from the same race and the third- plusgeneration native proximal host groups. Specifically, the host groups for Chinese, Cubans, Armenians, and Nigerians are non- Hispanic Asians, Hispanics, non- Hispanic whites, and non- Hispanic blacks. The results from figure 2 show that the second generation from all four ethnic groups outperformed both their nativeborn proximal hosts and their secondgeneration nonethnic counterparts from the same racial group. The higher achievement among the second generation is a key consequence of hyper- selectivity among the first generation. The gap between the ethnic group and proximal host group is smallest among Chinese and largest among Nigerians. By illustration, second- generation Chinese achievement was similar to the average achievement or the norm for their proximal host group. By contrast, second- generation Nigerian achievement was an outlier for their proximal host. Among second- generation Asians, the gap between Chinese and non- Chinese is small but for the other groups quite significant. For example, second- generation Cubans were twice as likely to finish college than both Puerto Ricans and other second- generation Hispanics (that is, second- generation non- Cuban Hispanics). Second- generation Nigerians were five times as likely as African Americans and twice as likely as other second- generation blacks (that is, second- generation non- Nigerian blacks) to have a college degree. Multivariate Analyses Further results from our multivariate analyses confirm the patterns documented in our descriptive analyses. Table 2 presents multivariate results from logistic regressions predicting college attainment for the four second- generation groups in comparison with the three native proximal host groups. Model 1 shows that blacks and Puerto Ricans were about half as likely as whites to have a bachelor s degree and that the second generation from all four ethnic groups were significantly more likely have graduated from college than the first generation. The largest differences are among Nigerians, who were 5.7 times more likely than whites to have a college degree, and among Chinese, who were 3.2 times more likely. Controlling for age, gender, region, and survey year, these key differences persist and remain statistically significant in model 2. The inclusion of survey year as a control variable does not change our results in any substantive way, although the survey year variables are statistically significant. Figure 3 graphs the predicted probabilities from multivariate analyses for the key ethnic groups and reveal a clear second- generation advantage over the three proximal host groups. Table 3 focuses on interethnic and intergenerational comparisons within the same racial group. The independent variable of interest is the ethnic group by immigrant generation, with second- generation non- coethnic individuals from the same racial group (that is, non- Chinese Asians, non- Cuban Hispanics, and so on) as the reference group. In other words, we examine how well second- generation Chinese fared relative to second- generation non- Chinese Asians. If first- generation hyperselectivity matters, as we posit, we expect the second generation from such groups to fare better than their counterparts from the other ethnic groups within the same U.S. racial group. We realize that this approach is imperfect, in that non- Chinese Asians still lump together diverse Asian ethnic groups, including high- achieving ones such as Asian Indians and Koreans as well as low- achieving ones such as Cambodians and Laotians. Compared with other second- generation Asians, the second- generation Chinese reported significantly higher odds of having a bachelor s degree or more. They were also significantly more likely than first- generation Chinese to achieve more education. Among Hispanics, both first- and second- generation Cubans were significantly more likely to complete a college education or more than other second- generation Hispanics. Nigerians were in fact the most highly educated group. Firstgeneration Nigerians were 2.5 times more likely and second- generation Nigerians 4.2 times more likely to have a bachelor s degree or higher than other second- generation blacks. Finally, first- generation Armenians were the only group that reported lower odds of having

13 the remaking of race 199 Table 2. Logistic Regression for Second-Generation Educational Attainment, College Graduates Variables Model 1 Model 2 Ethnoracial origin and immigrant generation Non-Hispanic black, third-plus generation 0.475*** 0.465*** (0.007) (0.007) Hispanic/Latino, third-plus generation 0.419*** 0.355*** (0.011) (0.009) Non-Hispanic Asian, third-plus generation 2.216*** 1.860*** (0.124) (0.105) Chinese, second generation 3.235*** 2.654*** (0.296) (0.235) Cuban, second generation 1.403** (0.166) (0.143) Armenian, second generation 2.787* 3.111** (1.195) (1.247) Nigerian, second generation 5.692*** 4.752*** (2.360) (1.988) Control variables Age 1.018*** (0.002) Age-square 1.000*** (0.000) Male (0.010) Midwest 0.720*** (0.011) South 0.802*** (0.011) West 1.048** (0.016) CPS 2010 vs. CPS ** (0.012) CPS 2012 vs. CPS *** (0.013) Constant 0.488*** 0.520*** (0.003) (0.028) N 285, ,238 Source: Authors compilation based on the CPS ASEC (Bureau of Labor Statistics 2012). Note: Odds ratios reported. Robust standard errors in parentheses. The reference group for ethnoracial origin is non-hispanic white, third-plus generation. The reference category for region is northeast. *p <.05; **p <.01; ***p <.001

14 200 Immigration and Changing Identities Figure 3. Predicted Probabilities of Educational Attainment by Ethnoracial Origin 1 Having a College Degree Predicted Probabilities White Black Hispanic Asian Chinese Cuban Armenian Nigerian Ethnoracial Origin Source: Authors compilation based on the CPS ASEC (Bureau of Labor Statistics 2012). Note: Combined sample is limited to population age twenty-five and older. The four native (third-andhigher-generation) groups are non-hispanic white, non-hispanic black, Hispanic and non-hispanic Asian. The four second-generation groups are Chinese, Cuban, Armenian, and Nigerian. Predicted probabilities are based on multivariate models, which also controlled for gender, age, quadratic tem of age, region, and survey year, holding these control variables at mean value. a bachelor s degree or more than other secondgeneration whites, whereas second- generation Armenians were 2.5 times more likely to have completed a college education. Figure 4 graphs the race- specific predicted probabilities from the multivariate analyses in table 3, holding all the control variables at their mean value. Except for Asians, the other three racial groups show three clear patterns. First, there is upward mobility in educational attainment when comparing first- and secondgeneration respondents within each ethnic group. Second, second- generation respondents from the hyper- selected groups also reported significantly higher achievements than their second- generation nonethnic counterparts from the same racial group. Third, among Asian respondents, second- generation Chinese respondents slightly outperformed first- generation Chinese and other second- generation Asians. Hyper-Sselectivity and the Cognitive Construction of Race Our analyses point to the positive association between hyper- selectivity and secondgeneration educational attainment. The hyperselectivity of first- generation Chinese, Cubans, Nigerians, and Armenians has led to even higher college completion rates among the second. Although hyper- selectivity positively affects college graduation rates for the second generation, what remains to be seen is whether this advantage will last beyond the second generation. Drawing from research on immigration and race- ethnicity, we considered how hyperselectivity might affect third- and latergeneration Chinese, Cubans, Nigerians, and Armenians, and theorize what this suggests about the effects of hyper- selectivity on the cognitive construction of race and patterns of eth-

15 the remaking of race 201 Table 3. Race-Specific Logistic Regression for Second-Generation Educational Attainment Variables Chinese Model 1 Cuban Model 2 Armenian Model 3 Nigerian Model 4 Immigrant generation First generation 1.215* 1.412*** 0.592* 2.489*** (0.096) (0.112) (0.129) (0.436) Second generation 1.441*** 2.668*** 2.462* 4.210*** (0.158) (0.335) (1.064) (1.773) Control variables Age *** 1.064*** 1.094* (0.013) (0.013) (0.011) (0.039) Age-square 1.000*** 0.999*** 0.999*** 0.999** (0.000) (0.000) (0.000) (0.000) Male 1.141* 0.738*** 1.258*** (0.071) (0.042) (0.066) (0.158) Midwest 1.978*** * (0.236) (0.115) (0.062) (0.350) South 1.557*** 0.843* (0.153) (0.072) (0.084) (0.206) West 1.356*** 0.768** (0.103) (0.066) (0.072) (0.248) CPS 2010 vs. CPS (0.082) (0.071) (0.065) (0.162) CPS 2012 vs. CPS ** (0.076) (0.071) (0.078) (0.228) Constant *** 0.248*** 0.092** (0.361) (0.024) (0.069) (0.075) N 5,709 9,214 9, Source: Authors compilation based on the CPS ASEC (Bureau of Labor Statistics 2012). Note: Odds ratios reported. Robust standard errors in parentheses. The reference group for each race-specific regression is second-generation individuals in same racial group, excluding the ethnic group. For Chinese, it is second-generation non-chinese Asian. For Cubans, it is second-generation non-cuban Hispanic. For Armenians, it is second-generation non-armenian white. For Nigerians, it is second-generation non-nigerian black. The reference category for region is northeast. *p <.05; **p <.01; ***p <.001 noracial identification. We contend that the effects of hyper- selectivity differ for groups depending on how they are racialized in the U.S. context, as well as the status of the proximal host group in relation to the hyper- selected immigrant group. Members of the second- generation may identify by national origin or ethnicity, and enter U.S. institutions (such as schools) with entrenched racial categories and highly stratified racial hierarchies. These, in turn, affect how teachers, guidance counselors, and peers perceive, treat, and identify students of diverse immigrant and ethnoracial backgrounds (Calarco 2014; Drake 2017; Ferguson 2003; Lee and Zhou 2015; Lewis- McCoy 2014; Valenzuela 1999). For example, although second- generation Nigerian students may strongly identify with their ethnicity and immigrant origin, school officials may recognize neither; rather, they may identify the students as black and treat them accordingly.

16 202 Immigration and Changing Identities Figure 4. Race-Specific Predicted Probabilities of Educational Attainment by Generation 1 Non-Hispanic Asian 1 Hispanic/Latino Predicted Probabilities Predicted Probabilities Chinese (1st gen) Chinese (2nd gen) Other Asian (2nd gen).0 Cuban (1st gen) Cuban (2nd gen) Other Hispanic (2nd gen) 1 Non-Hispanic White 1 Non-Hispanic Black Predicted Probabilities Predicted Probabilities Armenian (1st gen) Armenian (2nd gen) Other white (2nd gen) Nigerian (1st gen) Nigerian (2nd gen) Other Black (2nd gen) Source: Authors compilation based on the CPS ASEC (Bureau of Labor Statistics 2012). Note: Combined sample is limited to population age twenty-five and older. Other Asian includes non- Chinese Asians; Other Hispanic includes non-cuban Hispanics; Other White includes non-armenian whites; Other Nigerian includes non-nigerian blacks. Predicted probabilities are based on multivariate models, which also controlled for gender, age, quadratic term of age, region, and survey year, holding these control variables at mean value. Research shows that internal ethnic identification among the second generation can be at odds with external racial ascription, and studies of the children of black immigrants underscore the relevant difference between the two (Imoagene 2017; Tran 2015; Waters 1999). In her study of second- generation Nigerians, Imoagene shows that they succeeded in part by actively choosing their ethnicity while negotiating their race (2017). Despite their extraordinary academic achievement, however, she also finds that because of their racial status as black, they faced biases and barriers that impeded their full integration into U.S. institutions (see also Owens and Lynch 2012; Owens and Massey 2011; Patacchini and Zenou 2016). Like Van Tran and Mary Waters, in their studies of second- generation West Indians, Imoagene cautions that the class and ethnic advantages of the second- generation Nigerians may not extend to the third and later generations because of their racial status and the cognitive construction of blackness in U.S. society (Imoagene 2017; Tran 2015; Waters 1999). Here, we note that the size of the hyperselected immigrant group in relation to their proximal host matters for changing the cognitive construction of race, and has implications for ethnoracial identification among descendants of immigrants. For example, in spite of the hyper- selectivity of Nigerian immigrants and the extraordinarily high level of education attained by the second generation, Nigerians make up only 1 percent of the total U.S. black population (see table 4). This fraction is not enough to change the cognitive construction of blackness, which was born out of the legacy of slavery, entrenched by Jim Crow laws, and